Skip Comparing Uniform Window Results in Var/std Tests

[isVoid]

Close #9404

The issue in 9404 results from that pandas computes different rolling std results on x86 compared to aarch64 on a uniform window. In this specific test case, the window is

(Pdb) pdf['1'][574:576]
574    2.064967e+153
575    2.064967e+153
Name: 1, dtype: float64

On aarch64, pandas computes

(Pdb) mask = np.isclose(expect['1'].fillna(0), got['1'].fillna(0).to_pandas())
(Pdb) expect['1'][~mask]
575    1.149735e+145

On x86, pandas computes

(Pdb) expect['1'][575]
0.0

Since the results computed on uniform windows by pandas is not reliable, this PR drops these rows before comparing the results.

[codecov]

Codecov Report

Merging #9416 (d105746) into branch-21.12 (442009b) will decrease coverage by 0.05%.
The diff coverage is 0.00%.

❗ Current head d105746 differs from pull request most recent head 6830ace. Consider uploading reports for the commit 6830ace to get more accurate results

@@               Coverage Diff                @@
##           branch-21.12    #9416      +/-   ##
================================================
- Coverage         10.79%   10.74%   -0.06%     
================================================
  Files               116      116              
  Lines             19326    19506     +180     
================================================
+ Hits               2087     2096       +9     
- Misses            17239    17410     +171     

Impacted Files	Coverage Δ
python/cudf/cudf/_lib/__init__.py	0.00% <ø> (ø)
python/cudf/cudf/_lib/strings/__init__.py	0.00% <0.00%> (ø)
python/cudf/cudf/core/_base_index.py	0.00% <0.00%> (ø)
python/cudf/cudf/core/column/categorical.py	0.00% <0.00%> (ø)
python/cudf/cudf/core/column/column.py	0.00% <0.00%> (ø)
python/cudf/cudf/core/column/datetime.py	0.00% <0.00%> (ø)
python/cudf/cudf/core/column/lists.py	0.00% <0.00%> (ø)
python/cudf/cudf/core/column/numerical.py	0.00% <0.00%> (ø)
python/cudf/cudf/core/column/string.py	0.00% <0.00%> (ø)
python/cudf/cudf/core/column/struct.py	0.00% <0.00%> (ø)
... and 24 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 442009b...6830ace. Read the comment docs.

[shwina]

@gpucibot merge